摘要
研究目的:铁路工程项目前期投资估算的准确性对铁路建设项目的方案比选和投资控制起着至关重要的作用。目前,我国铁路工程投资估算主要以单位指标法、概预算定额法为主,整个过程相对复杂且耗时,估算准确性也依赖于从业人员的工作经验和能力水平。本文利用机器学习算法,以铁路隧道洞身工程为研究对象,构建基于支持向量机(SVM)和极限学习机(ELM)的造价预测模型,并收集若干隧道样本对模型进行训练和测试。通过实例验证和比较,选择适用性更强的算法,利用历史数据快速对隧道洞身造价进行预测,从而提高投资估算和方案比选的精度和效率。研究结论:(1)相较于ELM算法,SVM具有较高的预测精度和稳定性,预测结果的平均绝对百分比误差(MAPE)仅为3.41%,满足精度要求;(2)本研究成果为铁路隧道洞身工程的造价评估和预测提供了一种新型的、智能的数据驱动型的建模方法,并且通过仿真结果可知,该模型具有较好的可行性和适用性。
Research purposes: The accuracy of the preliminary investment estimation of railway engineering projects plays a vital role in the comparison and selection of railway construction projects and investment control. At present, our country’s railway engineering investment estimation is mainly carried out by the unit index method and the budget quota method. However, the whole process is relatively complicated and time-consuming, and the accuracy of the estimation also depends on the working experience and ability of the practitioners. In this paper, machine learning algorithms is used, main tunnel in railway tunnel engineering as the research object, to build a cost prediction model based on support vector machine(SVM) and extreme learning machine(ELM), and collect several tunnel samples to train and test the model. Through actual verification and comparison, a more applicable algorithm is selected, which can use historical data to quickly predict the cost of the main tunnel, thereby improving investment estimation and program comparison precision and efficiency.Research conclusions:(1) The result shows that SVM has high prediction accuracy and stability, compared with the ELM algorithm, and the mean absolute percentage error(MAPE) of the prediction results is only 3.41%, which meets the accuracy requirements.(2) The research results can provide a new and intelligent data-driven modeling method for the cost evaluation and prediction of main tunnel in railway tunnel engineering, and the simulation results show that the model has good feasibility and applicability.
作者
刘少非
侯大山
LIU Shaofei;HOU Dashan(China Railway Engineering Design and Consulting Group Co.Ltd,Beijing 100055,China;China Academy of Railway Sciences Corporation Limited,Beijing 100081,China)
出处
《铁道工程学报》
EI
北大核心
2022年第5期108-113,共6页
Journal of Railway Engineering Society
基金
国家重点研发计划“科技冬奥”重点专项
京张高铁智能化服务关键技术与示范(2020YFF0304100)。
关键词
支持向量机
极限学习机
铁路隧道
造价
预测
support vector machine
extreme learning machine
railway tunnel
construction cost
prediction
